From d86fb9f87a3be1bbeef6aecfb109e48d139413dc Mon Sep 17 00:00:00 2001
From: Jon Siwek <jsiwek@corelight.com>
Date: Fri, 18 Jan 2019 20:54:22 -0600
Subject: [PATCH] GH-151: fix hash calculation for nested sets

Hash key construction of nested sets depended on the order in
which their elements are iterated, which varied even between sets
containing equivalent elements.  The iteration order is now sorted
by each element's hash value (or, on collision, by full key) such
that equivalent sets no longer hash differently.
---
 src/CompHash.cc                               | 41 +++++++++++++++++--
 .../output                                    | 14 +++----
 testing/btest/language/nested-sets.bro        | 19 +++++++++
 3 files changed, 64 insertions(+), 10 deletions(-)
 create mode 100644 testing/btest/language/nested-sets.bro

diff --git a/src/CompHash.cc b/src/CompHash.cc
index f120c3618b..81ff353ab8 100644
--- a/src/CompHash.cc
+++ b/src/CompHash.cc
@@ -7,6 +7,9 @@
 #include "Reporter.h"
 #include "Func.h"
 
+#include <vector>
+#include <map>
+
 CompositeHash::CompositeHash(TypeList* composite_type)
 	{
 	type = composite_type;
@@ -174,12 +177,44 @@ char* CompositeHash::SingleValHash(int type_check, char* kp0,
 			{
 			int* kp = AlignAndPadType<int>(kp0);
 			TableVal* tv = v->AsTableVal();
-			ListVal* lv = tv->ConvertToList();
 			*kp = tv->Size();
 			kp1 = reinterpret_cast<char*>(kp+1);
-			for ( int i = 0; i < tv->Size(); ++i )
+
+			auto tbl = tv->AsTable();
+			auto it = tbl->InitForIteration();
+			ListVal* lv = new ListVal(TYPE_ANY);
+
+			struct HashKeyComparer {
+				bool operator()(const HashKey* a, const HashKey* b)
+					{
+					if ( a->Hash() != b->Hash() )
+						return a->Hash() < b->Hash();
+					if ( a->Size() != b->Size() )
+						return a->Size() < b->Size();
+					return strncmp(static_cast<const char*>(a->Key()),
+					               static_cast<const char*>(b->Key()),
+					               a->Size()) < 0;
+					}
+			};
+
+			std::map<HashKey*, int, HashKeyComparer> hashkeys;
+			HashKey* k;
+			auto idx = 0;
+
+			while ( tbl->NextEntry(k, it) )
 				{
-				Val* key = lv->Index(i);
+				hashkeys[k] = idx++;
+				lv->Append(tv->RecoverIndex(k));
+				}
+
+			for ( auto& kv : hashkeys )
+				delete kv.first;
+
+			for ( auto& kv : hashkeys )
+				{
+				auto idx = kv.second;
+				Val* key = lv->Index(idx);
+
 				if ( ! (kp1 = SingleValHash(type_check, kp1, key->Type(), key,
 				                            false)) )
 					{
diff --git a/testing/btest/Baseline/language.record-index-complex-fields/output b/testing/btest/Baseline/language.record-index-complex-fields/output
index f47493ea82..c07d86ee5e 100644
--- a/testing/btest/Baseline/language.record-index-complex-fields/output
+++ b/testing/btest/Baseline/language.record-index-complex-fields/output
@@ -10,18 +10,18 @@ a
 }
 }
 {
-[a=4, tags_v=[0, 1], tags_t={
-[two] = 2,
-[one] = 1
-}, tags_s={
-b,
-a
-}],
 [a=13, tags_v=[, , 2, 3], tags_t={
 [five] = 5,
 [four] = 4
 }, tags_s={
 c,
 d
+}],
+[a=4, tags_v=[0, 1], tags_t={
+[two] = 2,
+[one] = 1
+}, tags_s={
+b,
+a
 }]
 }
diff --git a/testing/btest/language/nested-sets.bro b/testing/btest/language/nested-sets.bro
new file mode 100644
index 0000000000..e33e1ac842
--- /dev/null
+++ b/testing/btest/language/nested-sets.bro
@@ -0,0 +1,19 @@
+# @TEST-EXEC: for i in `seq 21`; do echo 0 >> random.seed; done
+# @TEST-EXEC: test `bro -b -G random.seed %INPUT` = "pass"
+
+type r: record {
+	b: set[count];
+};
+
+global foo: set[r];
+global bar = set(1,3,5);
+
+add foo[record($b=bar)];
+
+bar = set(5,3,1);
+delete foo[record($b=bar)];
+
+if ( |foo| > 0 )
+	print "fail";
+else
+	print "pass";